In Evanz's summary of last Friday's Sports Analytics Innovation Summit, he mentioned that Golden State Warriors assistant GM and Santa Cruz Warriors GM Kirk Lacob "bolted from the conference immediately after his talk, so I didn't get a chance to catch up with him." Coincidentally, Lacob was leaving that conference to get back to Santa Cruz to meet with me. So between the two of us, GSoM pretty much spent most of a day listening to Lacob.
Fittingly, Lacob, Sammy Gelfand and I discussed how the Golden State Warriors are using advanced statistics to help improve their performance at both the D-League and NBA level. Naturally, that led me to wonder how the Warriors' analytics guys work with head coach Mark Jackson given his lukewarm comments about advanced statistics during his introductory press conference.
Again, this was already pretty long so I've some parts of this that were redundant, but I think this not only offers insight into how the franchise thinks about numbers but also why communication is such an important leadership trait in translating advanced statistics into improved performance.
We begin right after he described which statistics transfer best from college to D-League to NBA, which I'll summarize: rebounding and, for guards especially, blocks and steals as a sign they can compete athletically.
- - -
Nate Parham: Do you guys use the numbers - like plus/minus lineup data - here (in Santa Cruz) actively? Or is it more just a scouting thing?
Kirk Lacob: We're gonna do a lot more. We did a little bit last year. We actually, especially after the season - again things moved so quickly last year that we didn't have time to set up all the systems we wanted to, but we started to put stuff (in place) during the year - we did a little analysis after the year. And heading into this year, I think, we're a lot more ready to take advantage of some of the things we've also tried out in the NBA.
We're going to try more things down here because we can. But we've got a lot of opportunity to try a lot of things we've wanted to try for two years and just kind of test 'em out. And Sammy is very, very proficient in all this stuff and does a lot of data work. And he works with Pat Sund, who does a lot of our data analysis on a day-to-day basis up there and they talk.
Sammy Gelfand: We talk probably everyday. Just what Pat uses, what I use, trying to get things to merge. You know, I did lineup analysis and one thing I always stress is there's so many players that come through the D-League, the minutes are kind of...
NP: The sample size...
SG: There are some lineups that just played one minute together. And then somebody got traded, we had some guys get called up, some went to Europe. But you know after talking with Kirk and Pat and the rest of our coaching staff, we were able to find stuff that the coaches didn't realize at the end of the year. So I'm looking forward to because of the proximity we can do a lot more of that stuff this year.
NP: And the Warriors were one of the teams that set up the arena with the new tracking system...
KL: Sports VU. It's funny you mention that because I actually talked at this sports analytics conference in San Francisco this morning and my presentation was based around new technology in data analysis that the Warriors are starting to use in the last two years, our general outlook and how it's changed due to ownership's mandates. And so Sammy and Pat, they were there the last two days. I was only there for my part but they stayed. It's funny because I mentioned, a very large part of my presentation was about that camera and what we're doing with it up there.
We've got some really, really exciting stuff starting to happen with that - I think, I could be wrong, but I'm pretty sure the only team who has figured out how to crack the algorithms to track plays.
SG: And if we're not, it's only one of two. It's not many.
KL: As an example, Houston is probably at the forefront of almost everything analytically. When they heard about what we've been able to do over the summer, they were talking to our guys and, 'How did you do that? We ran into this problem..' I mean, to hear them - we're talking Daryl Morey and Sam Hinke - saying, 'How did you do this? We ran into this wall...' And our guy just said, 'I'm not gonna tell you - I'm not going to solve it for you. This is worth a lot of money!'
To hear them say that says that I think we're starting to push the envelope. I'm not ever saying we're the latest and greatest and we're ahead of everybody. But we are really in an advantageous position considering where we're located to have the resources of all these tech companies, all these entrepreneurial young guys.
I mean, this group that we have working on the data is a bunch of Cal Tech and Stanford grads. One of them is an aeronautical engineer person - aerospace engineer - and has done helicopter data analytics modeling, you know? I mean, these are guys that can program anything in the world; we've got guys who have masters or PhDs in computer science. I mean, these guys in an hour can whip up whatever.
So we're very fortunate because they're in our backyard and these are people that I have access to because they're friends of mine or they're friends of my friends. And there's very few other parts of the country where you can call a guy up and say, 'Hey, can you come by the office tomorrow and walk through this with us and show us what you've done? And we're going to give you some data and see what you can do with it.' So we're really fortunate to have that. And you know we just have to take advantage of it and in a way that makes sense.
NP: I assume you've seen Moneyball or read the book or both.
NP: Some people say that basketball has too many interaction effects on the court to really apply that to basketball. But what elements of Moneyball do you see working in basketball?
KL: It's funny because, again, I talked about this today.
NP: Well, sorry I missed it.
KL: It is really different. I grew up in the Bay Area, as I said, and I was always more of an A's fan than a Giants fan to be honest because my dad has known Billy Beane for a long time. So we were all for it during those runs in the early 2000's. And I grew up around technology and innovation and things like that. So I looked at it always as though that's pretty cool.
Baseball is really different. I mean, there are distinct events that are happening. You've got one pitcher throws a pitch - a certain type of pitch - you can log to one batter. And if he hits it, he hits it into a field that can easily be gridded. Basketball you got five on five and they're [in motion]. And even in football - we had a guy come talk today about football - it's got its own set of challenges in that it's 11 on 11 and there's so many little tiny things that matter so much because it's so quick; everything happens so quickly. At least for them, a football game is actually like 15 minutes long in terms of actual action. It is go-go-go-go, stop.
Basketball is tougher in a lot of ways because there's so much stuff that's changing all the time and there's no distinct events. That being said, the breakthrough is not going to be numbers-based, it's gonna be data-based. And that's where these cameras become so valuable because they log anything and everything that's on the court from if a guy's shot gets two degrees flatter from the first to fourth quarter, (we can ask) is that something that happens consistently? If this guy comes around a screen, he shoots five percent better when he creates that extra foot and a half of separation because so-and-so set a screen instead of this guy. This play works when we run it to this side and we create this space on average 10% better. This guy moves from this block to this block on weakside help five hundredths of a second quicker and that's the difference between tipping a ball and missing it.
Those are the type of things that will matter long-term. It's never going to be as simple as baseball where you can come up with WAR, VORP, all these statistics that are just based off of statistics that already existed. You know they have some new things - pitch effects, field effects. Basketball, it's going to be more difficult, until technology catches up.
Now luckily, I don't know how much you know about Sports VU cameras, that technology comes from missile tracking technology.
NP: Yeah, I've read about it.
KL: Which is pretty advanced to begin with - so, I mean, I can't even begin to really comprehend. I understand more or less how it works - no way I could ever tell you [specifics]. I'm not a mechanical engineer or anything so I can't do it. But that's where the difference is going to be: it's the data instead of the numbers themselves.
NP: So how would you communicate - like you were talking about the space and separation off a screen and such - how would you communicate that to players, who maybe aren't used to thinking the game that way?
KL: That is as difficult as solving the data problem in some ways. The communication to coaches and then the communication from coaches to player. Obviously coaches are very talented at communicating things to players. It really helps when you have a guy like Mark Jackson, who has built so much equity with players by having been a player himself - they respect everything he says.
Very rarely am I going to go up to a player and say, 'Hey, you're shooting at a two degree lower angle in the game.' No, I'm not going to say that from the guy. I might say, 'Hey Klay, your shot is getting flatter at the end of games. You might want to work on those legs.' Or we might say, 'You're losing lift at the ends of games, that's why your shooting percentages are going down.' He'll understand that because he's missing shots, but he might not understand necessarily why he's missing shots. So we can tell him, 'Your legs are tired. You need to work on that.' And that will be great once we can fit in load capacities and all sort of things that they're working on.
But the communication between front office, especially when it comes to more analytical types like Sammy or me, it's, 'We're gonna have to wait. Find a way to approach the coach. Every coach is different.' Luckily, we have a staff that buys into everything we're doing. They might not agree with every little thing, but they were hired because we think they've got really good basketball minds but also they understand our culture and what we're trying to do and they're open to new ideas. And they say, 'You know what? We will listen to everything you bring to us. We might not agree every time, but we will listen and we're willing to talk about it.'
So you can't go too high level or people just start [shutting down] - they're like, 'I can't understand that right now. Don't tell me about...'
NP: Points per possession?
SG: One hundredth of a second of a rotation of a...
KL: The angle of this! Or the cosine of that! 'I haven't taken like algebra or geometry since I was in high school! Why are you telling me all these things?'
But they can understand, 'Hey this play works better when our spacing is a little better. Gotta work on our guys' spacing. That's why that play is not working - because our spacing is off. And we can show you: this is how you drew it up and this is the way they're running it. It might have seemed like a good idea to have so-and-so running off this screen, but guess what? We looked at the numbers: he just never creates separation. He's not quick enough. He does not have a good enough first step or he doesn't come tight enough around screens. Or it's because this guy sets awful screens - look at all the plays we found where this guy sets really awful screens.'
So that's the type of stuff they will [respond to]; over time, we figure out what doses, what type exactly, how do you say it to them. It certainly helps when they like you. Like players tend to respond well to me, they respond well to Sammy - that's really helpful. If you can get a player to play seven-up with you - or horse - after practice, then you know that they're going to listen to what you have to say because they at least have the respect with you to hang out with you.
SG: I mean, just to kind of piggyback on that, Casey and me had- coming into our playoff game against Bakersfield we kind of did some more advanced shot charts, advanced stats. And we went to the coaching staff and eventually word got around to the players because they'd hear us talking about it and Stef (Hannah), Mo (Baker), I think Leo (Lyons) - about four or five guys - came up to us after practice to see what we did.
So I think that when they know we're putting the work into it, they want to know about it. Obviously, each player is going to talk to you different because each player responds differently. But I think once the players start hearing about all the info, they're going to want to know it and they're going to want everything that gives them a heads up. So I think they're going to be very receptive to it once they - obviously we have to work on finalizing it and actually making sure it's right and how we want to present it but - I think they're really going to relish a lot of the information.
KL: The other thing is that players listen to players - they're the greatest reference. So if you can get one guy that matters. Like, again with the D-League thing, if you get Jeremy Lin going out and saying, 'The time in the D-League for me was crucial' everyone else says, 'I can be the next Jeremy Lin - the D-League is cool with me.' All it takes is one guy; that's why I said the Lamar Odom thing would've been so big. 'Oh wow, Lamar is doing the D-League for rehab - I can do it. This guy has been an All-Star. He's won championships.'
If it's a leader on the team- if Steph Curry does anything, the rest of players on our team are going to be like, 'I'm going to fall in line with Steph.'
NP: So he's on board with listening to you guys right now?
KL: Steph is a great guy, fantastic guy. And he's really, really intelligent. So he's really into listening to anything, anything we bring to him. And you know, just like anyone else, there's going to be a point of overload. But he's a great guy to test things with because he will - at the very least out of respect - he will listen to what you're saying. He will never blow you off. And at the best, he'll take it in and use it. And so when he then says, 'Yeah, I talked to coach' or 'I talked to Sammy and Pat about this, yeah, it helps. I realized I'm way better coming off shooting off of two dribbles, left-right - I hit a lot more shots. They're right - I am way better that way. That's crazy - I'm going to start to set myself up to get two dribbles left-right.'
So influential players are important. I guarantee you if LeBron James at any point came out and said, 'Man the D-League is great.' or 'These advanced statistics are awesome' everybody would be all over them. You know, it's game over.
NP: Actually when Mark Jackson was hired - I think at his press conference - someone asked him about stats and he was somewhat dismissive. Almost to the point of saying, 'I don't believe in stats.'* Did you encounter that attitude at the beginning? Has he changed? Or was that just overstated to begin with?
KL: I think that the way maybe the question was asked was about specifically advanced statistics. When you say that word, he as a former player is gonna be like, 'It's B.S. - points, rebounds, and assists.'
What he does understand is trying new things. Like again, if you give him a piece of paper and it says this is the best player on the free agent market or this is the third best player on our team based on his WAR rating. He's going to be like, 'No. No. I'd rather rely on what I've seen.' But if you've giving him things like this guy is shooting better coming from this side, that is advanced statistics - that's analytics. That he does listen to.
So I think, yeah, it was overstated. And I think he was kind of making the point, 'Nah, I'm not into this Moneyball geekiness.' But if there's something real out there then yeah.
*Corrective note: It's an exaggeration to say Jackson said he doesn't believe in stats - what he did say is that he has more faith in his experience than the numbers. Click here to listen to what - and how - he actually said about advanced statistics.